Professional Documents
Culture Documents
Multicomponente Flash Algorithm
Multicomponente Flash Algorithm
Multicomponente Flash Algorithm
Introduction
Most industrially produced polymers are mixtures of poly- Heidemann Ž1999. applied this approach to develop an algo-
mer components with varying chain length, chain chemical rithm for determining the cloud- and shadow-point curves of
compositions, degrees of branching, and so on. In other words, polydisperse polymer solutions. Another approach is to treat
polymers are polydisperse. This polydisperse nature of poly- the distribution with a set of discrete pseudocomponents
mers has a strong effect on the phase behavior of mixtures ŽChen et al., 1993; Sadowski and Arlt, 1995; Folie, 1996;
containing polymers ŽShultz and Flory, 1952; Koningveld and Behme et al., 1999; Cheluget et al., 2002; Jog and Chapman,
Staverman, 1968; Bonner et al., 1974; Folie, 1996; Bokis et 2002.. For unimodal molecular-weight distribution curves,
al., 1999.. For example, the molecular-weight distributions of pseudocomponents are generated by matching the statistical
polymers over different phases will be different from that of moments of the distribution to those of the pseudocompo-
the feed polymer being flashed. nents to be determined ŽTork et al., 1999. or by discretizing a
In executing polymer flash calculation, there are two dif- continuous distribution function using Gaussian quadrature
ferent approaches to account for the molecular-weight distri- method ŽCotterman et al., 1985, 1986; Cotterman and Praus-
bution ŽMWD. of polydisperse polymers. One is to treat the nitz, 1985.. These two methods, however, are limited to flash
distribution with continuous distribution functions ŽKehlen calculations with small numbers of pseudocomponents when
and Raetzsch, 1980; Gualtieri et al., 1982; Raetzsch and detailed shape of a molecular-weight distribution is not im-
Kehlen, 1983, 1985., such as Schulz᎐Flory distribution or portant.
Wesslau distribution, and to apply the functional theory to Many industrial polymerization processes involve multiple
derive an expression for the chemical potential. Phoenix and reactors in series or use multisite catalysts, resulting in poly-
mers with bimodal or multimodal MWD curves ŽHungenberg
et al., 2001; Cheluget et al., 2002.. Representation of these
Correspondence concerning this article should be addressed to C.-C. Chen.
Current addresses of: S. Behme, ZT-ENG-Healthcare Prozessplanung, Bayer multimodal MWD curves with continuous distribution func-
Corp., B58-2, 800 Dwight Way, Berkeley, CA 94710; Gabriele Sadowski, Lehrstuhl tions would be extremely cumbersome, if not impossible. In
¨ Thermodynamik, Universitat
fur ¨ Dortmund, Emil-Figge-Strasse 70, 44227 Dort-
mund, Germany. contrast, the pseudocomponent approach is preferred be-
sequently, the derived natural logarithm of the fugacity coef- nent p: r p, j Ž js1,2, . . . , N ..
ficient of a pseudocomponent in the mixture is a linear func- 䢇 Average segment number of pseudocomponents for
tion of its segment number and is independent of mole frac- component p: r p .
tions of pseudocomponents representing the polydisperse The following relations and summations hold for these
polymer. This leads to a highly efficient flash algorithm as quantities:
the computation time becomes independent of the number of
pseudocomponents used to represent the polymer MWD K
curve. Therefore, POLYMIX achieves the computational ef- Ý x i s1 Ž3.
ficiency required to perform flash calculations with polydis- is 1
N where R is the gas constant. Using Eq. 10 for Z, ares can also
rp s Ý X j r p, j Ž6. be calculated using the mole fraction of the polydisperse
js 1 polymer in the mixture and the average segment number of
pseudocomponents for the polydisperse polymer
x p, j s x p X j js1,2, . . . , N Ž7.
N
ares s ares Ž T , , x p ,r p ,... . . Ž 13.
x p rp s Ý x p, j r p , j . Ž8.
js 1
Here Z is the compressibility factor for the mixture, T is the Equation 14 applies to both monodisperse components and
temperature, and is the molar density. In Eq. 9, x p, j 4 and pseudocomponents in the polymer mixture. The residual
r p, j 4 are vectors of mole fractions and segment numbers of chemical potential, res
i , is given by
pseudocomponents in the mixture, respectively, as defined in
Eqs. 1, 4, 5 and 7. All other parameters and compositions for
⭸ ares ⭸ ares ⭸ ares
res q
monodisperse components are not explicitly displayed in Eq.
9 because they are irrelevant to this discussion.
i sa
res
⭸
q
⭸ xi
y Ý xl
ž / ⭸ xl
Since the segment size and energy parameters for all pseu-
docomponents of the polydisperse polymer are the same and ⭸ ares ⭸ ares
the only differences among the pseudocomponents are their
s ares q RT Ž Zy1 . q
⭸ xi
y Ý ž /
xl
⭸ xl
, Ž 15 .
segment numbers, the compressibility factor for the mixture
can be calculated using the mole fraction of the polydisperse
polymer in the mixture and the average segment number of where ⭸ aresr⭸ x i is a partial derivative that is always done to
pseudocomponents for the polydisperse polymer in general. the mole fraction stated in the denominator, while all other
Therefore, we can rewrite Eq. 9 as follows mole fractions are considered constant. The summation term
in Eq. 15 is over both monodisperse components and pseudo-
components
Zs Z Ž T , , x p ,r p , . . . . , Ž 10.
⭸ ares K ⭸ ares N ⭸ ares
where x p is the mole fraction of the polydisperse polymer in
the mixture and r p is the average segment number of pseu-
Ý xl ž /⭸ xl
s
i/ p
Ý xi
ž / ⭸ xi
q Ý x p, j
js 1
ž /
⭸ x p, j
. Ž 16 .
where a is the molar Helmholtz energy of the mixture and ⭸ ares ⭸ ares ⭸ x p ⭸ ares ⭸ r p
aig is the molar Helmholtz energy of a mixture of ideal gases s q . Ž 17.
⭸ x p, j ⭸ xp ⭸ xp, j ⭸ rp ⭸ x p , j
at the same temperature, volume, and compositions. From
⭸ xp
s1 Ž 18.
⭸ x p, j
⭸ rp r p, j rp
s y Ž 19.
⭸ x p, j xp xp
Therefore, we have
N ⭸ ares ⭸ ares
Ý x p, j
js 1
ž / ⭸ x p, j
s
⭸ xp
. Ž 21.
Equation 20 indicates that the derivative for pseudocompo- Figure 1. Two-phase TP flash for a mixture containing a
nents in the mixture depends on the segment numbers of polydisperse polymer.
pseudocomponents in the mixture, but is independent of the
mole fractions of pseudocomponents within the polymer ex-
plicitly. Substituting Eq. 21 in Eq. 16, we obtain It should be pointed out that similar conclusions from Eqs.
23᎐25 were also obtained by Phoenix and Heidemann Ž1999.
⭸ ares K ⭸ ares ⭸ ares K ⭸ ares and Jog and Chapman Ž2002. in their formulations for bub-
Ý xl ž /⭸ xl
s Ý
i/ p
xi
ž /⭸ xi
q xp
⭸ xp
s Ý xi
is 1
ž / ⭸ xi
. ble-point calculations. However, unlike their approaches for
a specific EOS model in each case, we arrive at a general
Ž 22. conclusion that is independent of a specific segment-based
polymer EOS.
Applying Eqs. 20 and 22 in Eqs. 14 and 15, we can derive a
general expression for the fugacity coefficients of pseudo- TP Flash
components in the mixture In a two-phase temperature pressure ŽTP. flash calcula-
tion, temperature, pressure, and feed-stream compositions
ln p, j s ln A Ž T , , x p ,r p , . . . . q r p , j ln B Ž T , , x p ,r p , . . . . are defined. From a mathematical point of view, it is a prob-
lem of solving a system of coupled nonlinear equations de-
js1, 2, . . ., N, Ž 23. rived from phase equilibrium conditions and mass balances.
In the particular case of interest, in which one component is
where a polydisperse polymer, the flash calculation leads to a poly-
mer MWD in each phase that is different from the polymer
K ⭸ ares MWD of the feed stream, as shown in Figure 1. In this sec-
RT ln A s ares q RT w Ž Zy1 . yln Z x y Ý xi
is 1
ž / ⭸ xi tion, we derive all equations needed for the flash calculation
from phase equilibrium conditions and mass balances.
⭸ ares r p ⭸ ares
q y Ž 24.
⭸ xp x p ⭸ rp Phase equilibrium and mass balance for monodisperse
components
1 ⭸ ares
RT ln B s . Ž 25. For a Ž K y1q N . component system, the phase equilib-
x p ⭸ rp rium conditions for the K y1 monodisperse components in
the mixture are
Equations 23᎐25 show that the natural logarithm of the
fugacity coefficient of a pseudocomponent in the mixture is a 0 s x iI iI y x iII iII is1, . . ., K , i/ p, Ž 26.
linear function of its segment number and is independent of
mole fractions of pseudocomponents within the polymer. Also where I and II denote phase 1 and 2, respectively, and i is
note that both ln A and ln B are the same for all pseudo- the fugacity coefficient of component i in the mixture.
components. As we will show in the next section for flash Mass balances are also needed for flash calculations. The
algorithm formulation, Eqs. 23᎐25 are the key equations that mass-balance component for all monodisperse components in
lead to a highly efficient flash algorithm as the computation the mixture can be written as
time for polymer systems becomes independent of the num-
ber of pseudocomponents used to represent the polydisperse
polymer. n iF s nIi q nII
i is1, . . . , K , i/ p, Ž 27.
0 s x Ip, j pI , j y x II
p , j p , j
II
js1, 2, . . . , N, Ž 33. x pF N p,II j
y1
0 s1y Ý X jF ␣ II q ␣ I . Ž 42.
x II
p js 1 p,I j
where p, j is the fugacity coefficient of pseudocomponent j
in the mixture. Using Eq. 7, we can rewrite Eq. 33 as
Similarly, if we apply the same procedure for phase I, we can
0s x Ip X jIp,I j y x pII X jII pII, j js1, 2, . . . , N. obtain
Ž 34.
y1
The same mass balance equation, Eq. 27, for monodisperse x pF N p,I j
0 s1y Ý X jF ␣ q␣I II
. Ž 43.
components can also apply to pseudocomponents, leading to x Ip js 1 p,II j
x pF X jF s ␣ I x Ip X jI q ␣ II x II
p Xj
II
js1, 2, . . . , N. Ž 36. y1
x pF N p,II j
0 s r pII y Ý X jF r p, j ␣ II q ␣ I . Ž 44.
In addition, the summation of mole fractions of pseudocom- x II
p js 1 p,I j
ponents in each phase is also unity
Finally, if we multiply the segment number of pseudocom-
N
ponent j on both sides in Eq. 36, then apply Eq. 6 for the
Ý X jI s1 Ž 37. average segment number of pseudocomponents, we can ob-
js 1
tain the fourth equation
N
Ý X jII s1. Ž 38.
js 1 x pF r pF s ␣ I x Ip r pI q ␣ II x pII r pII . Ž 45.
K K
r pF x pF y ␣ II r pII x pII
Ý x iI y Ý x iII s 0. Ž 47. r pI s . Ž 54.
is 1 is1 ␣ I x Ip
Literature Cited
Behme, S., G. Sadowski, and W. Arlt, ‘‘Modeling of the Separation
of Polydisperse Polymer Systems by Compressed Gases,’’ Fluid
Phase Equilib., 158–160, 869 Ž1999..
Bokis, C. P., H. Orbey, and C.-C. Chen, ‘‘Properly Model Polymer
Processes,’’ Chem. Eng. Prog., 39 ŽApr. 1999..
Bonner, D. C., D. P. Maloney, and J. M. Prausnitz, ‘‘Calculation of
High-Pressure Phase Equilibria and Molecular Weight Distribu-
tion in Partial Decomposition of Polyethylene-Ethylene Mixture,’’
Ind. Eng. Chem. Process Des. De®., 13, 91 Ž1974..
Chapman, W. G., G. Jackson, and K. E. Gubbins, ‘‘Phase Equilibria
of Associating Fluids,’’ Mol. Phys., 65, 1057 Ž1988..
Chapman, W. G., G. Jackson, K. E. Gubbins, and M. Radosz, ‘‘New
Figure 5. Liquid –liquid equilibria of polypropylene Reference Equation of State for Associating Liquids,’’ Ind. Eng.
(PP) –propane at three temperatures. Chem. Res., 29, 1709 Ž1990..
Cheluget, E. L., C. P. Bokis, L. Wardhaugh, C.-C. Chen, and J. Fisher,
PC-SAFT-calculated isotherms are compared with experi- ‘‘Modeling Polyethylene Fractionation Using the Perturbed-Chain
mental cloud-point data of Whaley et al. Ž1997 ..
Statistical Associating Fluid Theory Equation of State,’’ Ind. Eng.
Chem. Res., 41, 968 Ž2002..
Chen, C.-K., M. A. Duran, and M. Radosz, ‘‘Two-Phase Equilibria in
Asymmetric Polymer Solutions; Block-Algebra, A New Flash Algo-
data were obtained by Whaley et al. Ž1997. for a weight-aver- rithm and SAFT Equation of State, Applied Single-Stage Dual
age molecular weight of Mw s 290 kgrmol and MwrMn s 4.4 Solvent Supercritical Fractionation of Polyethylene,’’ Ind. Eng.
Chem. Res., 32, 3123 Ž1993..
of polypropylene. Similar to the earlier correlation on Cotterman, R. L., R. Bender, and J. M. Prausnitz, ‘‘Phase Equilibria
PE᎐ethylene systems, a single constant value of 0.025 for k 12 for Mixtures Containing Very Many Components. Development
is used for the correlation at all three temperatures. Again, and Application of Continuous Thermodynamics for Chemical
the correlation results are in good agreement with the experi- Process Design,’’ Ind. Eng. Chem. Process Des. De®., 24, 194 Ž1985..
Cotterman, R. L., G. F. Chou, and J. M. Prausnitz, ‘‘Comments on
mental data.
‘Flash Calculations for Continuous or Semicontinuous Mixtures
Using an Equation of State’,’’ Ind. Eng. Chem. Process Des. De®.,
25, 840 Ž1986..
Cotterman, R. L., and J. M. Prausnitz, ‘‘Flash Calculations for Con-
Conclusions tinuous or Semicontinuous Mixtures Using an Equation of State,’’
Ind. Eng. Chem. Process Des. De®., 24, 434 Ž1985..
POLYMIX offers a computationally efficient algorithm for de Loos, Th. W., ‘‘Evenwichten tussen fluıde¨ fasen in systemen van
flash calculations of multicomponent systems containing linear polyetheen en etheen Žin Dutch.,’’ PhD Thesis, Delft, The
polydisperse polymers. Incorporating the concept of pseudo- Netherlands Ž1981..
components, the algorithm allows rigorous and generalized de Loos, Th. W., W. Poot, and G. A. M Diepen, ‘‘Fluid Phase Equi-
libria in the System PolyethyleneqEthylene. 1. Systems of Linear
flash calculations with multimodal polymer MWD. The algo- PolyethyleneqEthylene at High Pressure,’’ Macromolecules, 16, 111
rithm formulation is independent of the use of a specific seg- Ž1983..
ment-based polymer EOS for computing thermodynamic Folie, B., ‘‘Single-Stage Fractionation of PolyŽEthylene-co-Vinyl Ac-
properties of the polymer mixture. Furthermore, it takes ad- etate. in Supercritical Ethylene with SAFT,’’ AIChE J., 42, 3466
Ž1996..
vantage of the fact that polymer mixture properties Žsuch as
¨ einfache, assozi-
Gross, J., ‘‘Entwicklung einer Zustandsgleichung fur
pressure, compressibility factor, and Helmholtz energy. can ierende und makromolekulare Stoffe,’’ PhD Thesis, Technische
be calculated based on the number-average molecular weight Universitat¨ Berlin, Germany Ž2000..
of the polydisperse polymer. This leads to a highly efficient Gross, J., and G. Sadowski, ‘‘Perturbed-Chain SAFT: An Equation
and robust flash algorithm as the computation time for poly- of State Based on a Perturbation Theory for Chain Molecules,’’
Ind. Eng. Chem. Res., 40, 1244 Ž2001..
mer flash calculations becomes independent of the number Gross, J., and G. Sadowski, ‘‘Modeling Polymer Systems Using the
of pseudocomponents used to represent the polymer MWD Perturbed-Chain Statistical Associating Fluid Theory Equation of
curve. State,’’ Ind. Eng. Chem. Res., 41, 1084 Ž2002..